fitness.data <- read.csv('../data/compensation_4_30.csv', na.strings="NA")

fitness.data$prop.scd <- fitness.data$scd / (fitness.data$scd + fitness.data$wt)

rep.colors <- c("#EE2222", "#55AA00", "#1133FF", "#992299")

fitness.data$plot.col <- rep.colors[fitness.data$rep]

par(mar=c(4,4,1,1))

plot(x = fitness.data$gen, y = fitness.data$prop.scd, 
    pch = 16, col = fitness.data$plot.col, 
    xlab = "Generation", ylab = "Proportion of scd1 males")
    
for (r in 1:4) {
	cur.data <- fitness.data[fitness.data$rep == r,]
	lines(x=cur.data$gen, y=cur.data$prop.scd, col=cur.data$plot.col, lty=2)
}


# Note from ID, why not just fit the regression of frequency on generation?

library(sciplot)
ci.95 <- function(x) {
	mean.x <- mean(x)
	se.x <- se(x)
	return(c(mean.x-2*se.x, mean.x+2*se.x))	
}

pdf(file="../outputs/Figure4_fitness.pdf", width = 5, height = 4)

par(mar=c(5,5,1,1))
lineplot.CI(x.factor = fitness.data$gen, 
    response = fitness.data$prop.scd, 
    ci.fun = ci.95, xlab = "Generation",
    pch = 20, cex = 2, lwd = 2, cex.lab = 1.5,
    ylab = expression(paste("frequency of ", italic(scd)^1, " males")))
    #ylab = "Proportion of scd1 males")
dev.off()